WHERE short_char_column='literal' ignores the collation on a huge table

DROP TABLE IF EXISTS t1;

CREATE TABLE t1 (c CHAR(4)) ENGINE=MyISAM;

DELIMITER $$

FOR i IN 0..90000 DO

  INSERT INTO t1 VALUES ('aaaa');

  INSERT INTO t1 VALUES ('bbbb');

  INSERT INTO t1 VALUES ('AAAA');

  INSERT INTO t1 VALUES ('BBBB');

END FOR;

$$

DELIMITER ;

SELECT c INTO OUTFILE 't1.txt' FROM t1;

DROP TABLE IF EXISTS t2;

CREATE TABLE t2 (c CHAR(4) CHARACTER SET latin1 COLLATE latin1_bin) ENGINE=ColumnStore;

LOAD DATA INFILE 't1.txt' INTO TABLE t2;

SELECT * FROM t2 WHERE c='aaaa';

+------+

| c    |

+------+

| aaaa |

| aaaa |

| aaaa |

| aaaa |

| aaaa |

| aaaa |

| aaaa |

| aaaa |

| aaaa |

| aaaa |

........

| aaaa |

| AAAA |

| aaaa |

| AAAA |

+------+

147746 rows in set (0.054 sec)

SELECT DISTINCT c FROM t2 WHERE c='aaaa';

+------+

| c    |

+------+

| aaaa |

| AAAA |

+------+

DROP TABLE IF EXISTS t3;

CREATE TABLE t3 LIKE t2;

INSERT INTO t3 SELECT * FROM t2 WHERE c='aaaa';

SELECT c, COUNT(*) FROM t3 GROUP BY c;

+------+----------+

| c    | COUNT(*) |

+------+----------+

| AAAA |    57745 |

| aaaa |    90001 |

+------+----------+